Guest_mssql


Hello all,
I will try to supply all of the information available to me to help generate some answers to the question I have. I am doing some benchmark testing of an application of how it deals with various database queries, for example oracle 10g, DB2 and MSSQL 2005.
I have a very simple database called "student" that has various tables, with various columns, all as mentioned very simplistic. Each table has a few hundred rows of data, the most being one table with 2000 rows. The same database (structure and dataset), is replicated across all 3 DB's.
In order to do the testing, I have been recording the same queries (select * from table_a, select * from table_b etc etc) for each database at an interface which monitors traffic between the client and db server, the problem starts here; The size of the tcpdumps are vastly greater for MSSQL vs oracle or db2. Some examples:

100 query or transaction dump:
mssql=3.7mb
oracl10g=133kb

50,000 query or transaction dump:
mssql=1.8gb
oracl10g=0.6gb
db2=0.6gb

as mentioned each database's tables and data are identical to my knowledge and all queries are the same. The bloated dumps are making mssql performance numbers look bad.

So my question is: What could be the reason for MSSQLs client/server queries to contain so much information, has anyone else come across this Is there any setting or something I could try to minimize it

Thank you for your time.




Re: Number/size of packets at a network level for a MSSQL query Vs oracle vs DB2.

Robert Varga


Hello,

I'd be interested to know how you are issuing the statements. Are you issuing them in a uniform manner (i.e - all via ADO .net using OLEDB), or are you just executing them from the DB's respective client tool apps

Cheers,

Rob






Re: Number/size of packets at a network level for a MSSQL query Vs oracle vs DB2.

Guest_mssql

I have tried two methods for executing the queries:

1) In house java app that just uses a mssql driver to connect to the db, retrieves queries from a text file and executes them, can make setting adjustments like, ms between queries, re-connect after each query etc etc, I have made sure we do not reconnect and have everything as minimal as possible.

2) Using widely available management consoles, for example MSSQL server management studio express and oracle enterprise management console.

So I have used at least two different methods to execute the query, both methods have the same results, the mssql dump is larger.

Thanks for your reply so far.